Hi all,
I am a senior electrical engineering student, and have been reading posts on this forum regularly so I decided I would attempt to contribute to the community.
I recently implemented the Lm3875 ChipAmp kit from audiosector.com and decided to share some insight into how this thing works. The derivations you see apply to each of the two amplifier channels. The derivations are shown in the attached picture file.
This is intended for folks who know some basic circuit analysis techniques, but would still be a good thing to take a look at for anyone attempting to build this chipamp.
Notes:
for Voltage: dBV (or decibels referenced to a volt) = 20*log_10(Voltage (volts))
for Power: dBW (or decibels referenced to a watt) = 10*log_10(Power (watts))
subtracting dB values corresponds to dividing the original due to the properties of logarithms (adding dB values corresponds to multiplication of the original values.)
Gain values in dB are not referenced to a unit because units cancel when calculating gain.
for example Vout (volts)/Vin (volts) = voltage gain (unit-less or sometimes stated volts per volt).
Vout dBV-Vin dBV = Gain dB so when using dB values subtracting cancels units.
Voltage=Current*Resistance, Power = Voltage*Current =(Voltage^2)/Resistance. For AC voltage and current, V=Current*impedance
AC voltages and currents are sinusoidal with a frequency f. An audio signal consists of several sinusoids of varying frequency superimposed on each other.
V RMS stands for the Root mean square alternating voltage. This is used for AC measurements because the voltage and current are sinusoidal therefore being waves with peaks and troughs. VRMS=.707*Vpeak.
For an Ideal Op Amp, it is assumed that V+=V- and the current entering the + and - input terminals is 0. (this also implies that the input impedance is extremly high -> infinity, but in real world conditions it has a very large finite value.)
The || symbol rempresents a paralell combination of components. for resistors (or impedances Z) in paralell 1/Rtotal = 1/R1 + 1/R2 + 1/R3 ... ect for the case of two resistors in paralell this reduces to Rtotal = (R1*R2)/(R1+R2)
For capacitors the impedance Z = 1/j*(2*pi*f*C) so the impedance changes with frequency which is an over simplification of the concept filters are founded upon. j is the imaginary unit corresponding to sqaureroot(-1).
For AC frequencies not affected by the filter associated with a capacitor , the capacitance can be viewed as a short circuit (meaning that it is simply a wire). For DC, a capacitor can be represented as an open circuit (though there is a transient response to a sudden change in DC voltage).
The transfer function resulting from replacing R1 (on the audiosector board) with a capacitor corresponds to a high pass filter (because it is meant to block DC aka a frequency of 0.) fc in the attached picture corresponds to the -3dB frequency of the high pass filter transfer function.
This is a fancy way of saying that fc is the frequency at which the gain of the amplifier will have dropped by 3dB. This is important because fc= 1/(2*pi*C*R2) in this configuration which allows you to calculate the lowest capacitance that will not put fc anywhere near the lowest audio frequency.
(audio frequencies range from 20Hz to 20kHz).
The recommended 4.7uF from Peter Daniel's building instructions corresponds to fc=1.5Hz. Since this is a high pass filter, the audio frequencies are unaffected and as you get closer to 1.5Hz moving downard in frequency, the gain drops. At 1.5Hz the gain will have dropped by 3dB (corresponding to half power). The gain continues to drop until it is effectively zero at DC.
For the potentiometer representation it might be useful to note that the resistance between pin3 and pin 2 labeled as R32 becomes smaller while R21 becomes bigger for turning the knob clockwise, and the opposite is true for counter clockwise. Zo is the nominal impedance seen at V1 (the amplifier input). Because Zo is such a large value, Zo||R2 is approximately equal to R2.
The schematic showing the potentiometer corresponds to connecting it as shown with pin2 connected to the 4.7uF capacitor placed at R1. The capacitor is left out here because it is essentially a short circuit (direct connection) for audio frequencies when used with the component values recommended by Audiosector.com
If R1 is not replaced by a capacitor, refer to the following gain equation that includes R1. The high pass filter would not apply here and any DC offset would be amplified by (R2/(R1+R2))*(1+(RF/R3))
Typical mp3 player output is between 0.3V and 6V RMS depending on the volume level selected on the mp3 player.
The gain of this circuit is set to be about 30 which means that the output voltage = input voltage X 30.
The output voltage should not exceed 30V (peak) or the LM3875 chip will clip the output voltage.
The voltage supply rails for this kit depend on the transformer used. With the recommended transformer the will be approximately 30V and -30V. An output voltage greater than 30 will 'clip' the output waveform causing distortion.
in terms of power, the speaker is 8ohms so using the power equation, Power RMS=50=(Vout^2)/8. Solving for Vout (RMS) gives you 20V. since gain is 30, (20V output)/(30)=0.7 max input voltage. This gives you some margin before reaching an output voltage of 30V which corresponds to an input of 1V.
-John
I am a senior electrical engineering student, and have been reading posts on this forum regularly so I decided I would attempt to contribute to the community.
I recently implemented the Lm3875 ChipAmp kit from audiosector.com and decided to share some insight into how this thing works. The derivations you see apply to each of the two amplifier channels. The derivations are shown in the attached picture file.
This is intended for folks who know some basic circuit analysis techniques, but would still be a good thing to take a look at for anyone attempting to build this chipamp.
Notes:
for Voltage: dBV (or decibels referenced to a volt) = 20*log_10(Voltage (volts))
for Power: dBW (or decibels referenced to a watt) = 10*log_10(Power (watts))
subtracting dB values corresponds to dividing the original due to the properties of logarithms (adding dB values corresponds to multiplication of the original values.)
Gain values in dB are not referenced to a unit because units cancel when calculating gain.
for example Vout (volts)/Vin (volts) = voltage gain (unit-less or sometimes stated volts per volt).
Vout dBV-Vin dBV = Gain dB so when using dB values subtracting cancels units.
Voltage=Current*Resistance, Power = Voltage*Current =(Voltage^2)/Resistance. For AC voltage and current, V=Current*impedance
AC voltages and currents are sinusoidal with a frequency f. An audio signal consists of several sinusoids of varying frequency superimposed on each other.
V RMS stands for the Root mean square alternating voltage. This is used for AC measurements because the voltage and current are sinusoidal therefore being waves with peaks and troughs. VRMS=.707*Vpeak.
For an Ideal Op Amp, it is assumed that V+=V- and the current entering the + and - input terminals is 0. (this also implies that the input impedance is extremly high -> infinity, but in real world conditions it has a very large finite value.)
The || symbol rempresents a paralell combination of components. for resistors (or impedances Z) in paralell 1/Rtotal = 1/R1 + 1/R2 + 1/R3 ... ect for the case of two resistors in paralell this reduces to Rtotal = (R1*R2)/(R1+R2)
For capacitors the impedance Z = 1/j*(2*pi*f*C) so the impedance changes with frequency which is an over simplification of the concept filters are founded upon. j is the imaginary unit corresponding to sqaureroot(-1).
For AC frequencies not affected by the filter associated with a capacitor , the capacitance can be viewed as a short circuit (meaning that it is simply a wire). For DC, a capacitor can be represented as an open circuit (though there is a transient response to a sudden change in DC voltage).
The transfer function resulting from replacing R1 (on the audiosector board) with a capacitor corresponds to a high pass filter (because it is meant to block DC aka a frequency of 0.) fc in the attached picture corresponds to the -3dB frequency of the high pass filter transfer function.
This is a fancy way of saying that fc is the frequency at which the gain of the amplifier will have dropped by 3dB. This is important because fc= 1/(2*pi*C*R2) in this configuration which allows you to calculate the lowest capacitance that will not put fc anywhere near the lowest audio frequency.
(audio frequencies range from 20Hz to 20kHz).
The recommended 4.7uF from Peter Daniel's building instructions corresponds to fc=1.5Hz. Since this is a high pass filter, the audio frequencies are unaffected and as you get closer to 1.5Hz moving downard in frequency, the gain drops. At 1.5Hz the gain will have dropped by 3dB (corresponding to half power). The gain continues to drop until it is effectively zero at DC.
For the potentiometer representation it might be useful to note that the resistance between pin3 and pin 2 labeled as R32 becomes smaller while R21 becomes bigger for turning the knob clockwise, and the opposite is true for counter clockwise. Zo is the nominal impedance seen at V1 (the amplifier input). Because Zo is such a large value, Zo||R2 is approximately equal to R2.
The schematic showing the potentiometer corresponds to connecting it as shown with pin2 connected to the 4.7uF capacitor placed at R1. The capacitor is left out here because it is essentially a short circuit (direct connection) for audio frequencies when used with the component values recommended by Audiosector.com
If R1 is not replaced by a capacitor, refer to the following gain equation that includes R1. The high pass filter would not apply here and any DC offset would be amplified by (R2/(R1+R2))*(1+(RF/R3))
Typical mp3 player output is between 0.3V and 6V RMS depending on the volume level selected on the mp3 player.
The gain of this circuit is set to be about 30 which means that the output voltage = input voltage X 30.
The output voltage should not exceed 30V (peak) or the LM3875 chip will clip the output voltage.
The voltage supply rails for this kit depend on the transformer used. With the recommended transformer the will be approximately 30V and -30V. An output voltage greater than 30 will 'clip' the output waveform causing distortion.
in terms of power, the speaker is 8ohms so using the power equation, Power RMS=50=(Vout^2)/8. Solving for Vout (RMS) gives you 20V. since gain is 30, (20V output)/(30)=0.7 max input voltage. This gives you some margin before reaching an output voltage of 30V which corresponds to an input of 1V.
-John
Attachments
Last edited:
Gain is 30 because by experience the chip has the best sound around 26-30 gain. Not so keep the chip safe.
It wasn't my intention to make that point. You are correct, as the chip has internal protection circuitry to protect it if it gets too hot. There is however, a maximum amount of current that can flow through the output stage of this op-amp.
The output power is inherently dependent on the input signal level and gain.
a .7V input corresponds to 50W output for a gain value of 30.
a 1V input multiplied by 30 giving a 30V output would begin to clip (the power supply rails are set to about 30.5V for his kit.) So yes while it is true that it is intended to keep the amplifier from clipping, depending on the heat sink used and the input voltage level the device could overheat even with a gain of 30.
It would be nice to see some real world data on the output power for which the device will overheat and the internal protection circuitry will activate. It would still, of course, be dependent upon how well the device was heat sinked.
Okay, To clarify a few things:
From the LM3875 Data sheet
Absolute max output current rating: 6 Amps
Absolute Max power dissapation: 125 W
These ratings are absolute maximum, and the sound will have degraded far before these limits are reached.
These ratings also assume that the temperature of the chip does not exceed 25 degrees Celsius
The maximum dissipated power and average dissipated power calculations can be found in the datasheet in the thermal considerations section.
Thank you for your point about the gain. I have edited the original post to hopefully be more clear about this.
This post is more intended to provide a basic understanding of the circuit analysis involved with the non-Inverting op-amp configuration provided by Audiosector.com
From the LM3875 Data sheet
Absolute max output current rating: 6 Amps
Absolute Max power dissapation: 125 W
These ratings are absolute maximum, and the sound will have degraded far before these limits are reached.
These ratings also assume that the temperature of the chip does not exceed 25 degrees Celsius
The maximum dissipated power and average dissipated power calculations can be found in the datasheet in the thermal considerations section.
Thank you for your point about the gain. I have edited the original post to hopefully be more clear about this.
This post is more intended to provide a basic understanding of the circuit analysis involved with the non-Inverting op-amp configuration provided by Audiosector.com
Congratulations on using the correct model for predicting the volume control attenuation.
Vi(out) = Vin * [R21||Zo(in)] / [{R21||Zo(in)} + R32]
When R21 <<< Zin then Zin can be ignored with minor effect on the attenuation.
As Zin approaches 10*R21 the effect starts to become significant. When Rin = R21 the effect is gross.
I will read the rest later.
Vi(out) = Vin * [R21||Zo(in)] / [{R21||Zo(in)} + R32]
When R21 <<< Zin then Zin can be ignored with minor effect on the attenuation.
As Zin approaches 10*R21 the effect starts to become significant. When Rin = R21 the effect is gross.
I will read the rest later.
I got some 3875s without a purpose, but with some 3876/3886/4780 I'm at a loss what to do with them.
I was actually trying to figure out a polyphase motor drive using them, for my final year project ( assuming i get there)
For the output of a typical mp3 player, (say an eyepod) that the level isn't very high. Maybe 6 V p-p max. Mine is about half the level of my cd player.
I was actually trying to figure out a polyphase motor drive using them, for my final year project ( assuming i get there)
For the output of a typical mp3 player, (say an eyepod) that the level isn't very high. Maybe 6 V p-p max. Mine is about half the level of my cd player.
Last edited:
As R21 approaches Zin the term [R21||Zo(in)] has very little affect on the attenuation because the equation Vi(out) = Vin * [R21||Zo(in)] / [{R21||Zo(in)} + R32] has R21||Zo(in) in both the numerator and denominator.
Furthermore,
While R21 is becoming larger in value, R32 is becoming smaller, thus the scale factor [R21||Zo(in)] / [{R21||Zo(in)} + R32] approaches 1, therefore resulting in no input attenuation. This corresponds to turning the knob clockwise (turning the volume up)
For turning the knob counter clockwise, R21 will decrease and paralell combination R21||Zo = ((R21*Zo/(R21 + Zo)) quickly approaches zero because the numerator term R21*Zo will approach zero. As this takes place the scale factor [R21||Zo(in)] / [{R21||Zo(in)} + R32] is also quickly approaching zero thus resulting in severe attenuation.
Surely there is a safe operating area graph or calculator in the data sheet.
- Status
- Not open for further replies.